aug_gaussian_noise — Gaussian additive-noise augmentation

Group: Augmentation · Binding: n4m.sklearn.GaussianAdditiveNoise · C ABI: n4m_aug_gaussian_noise_*

Description

Add IID Gaussian noise to each element of X.

Parameters

Name

Type

Default

sigma

float

0.01

rng

Optional[PCG64]

None

seed

int

0

Explanations

Bibliographic source

Standard data-augmentation perturbation; see the nirs4all augmentation handbook.

Mathematical principle

Adds i.i.d. Gaussian noise \(\mathbf{x} + \boldsymbol{\varepsilon}\), \(\varepsilon \sim \mathcal{N}(0,\sigma^2)\), to each element, simulating detector/shot noise to regularise models and enlarge small calibration sets.

Implementation

C ABI n4m_aug_gaussian_noise_* in libn4m (create / apply / destroy lifecycle), wrapped by n4m.sklearn.GaussianAdditiveNoise. The same numerical kernel backs every language binding.

Usage

from n4m.sklearn import GaussianAdditiveNoise
op = GaussianAdditiveNoise()
X_transformed = op.fit_transform(X)

Benchmarks

Adaptive wall-clock per cell measured against full_matrix.csv. Only backends that implement this method are listed; libraries without the method are omitted.

Verdict  ·  ✓ ref / ≈ ref / ~ shape mark a reference-gate pass at strict / relaxed / qualitative tolerance  ·  ✓ bind = pls4all binding agrees with the C++ baseline  ·  ⇄ cross-check = documented by-design selector/RNG/model, noncanonical API/facade convention, or secondary oracle  ·  ✗ divergent  ·  ⚠ error  ·  — not run. The fastest backend per column is marked 🏆.

Reference gate: strict — numeric equivalence (rmse_rel_tol 1e-12).

BackendParity50×250 (ms)250×50 (ms)
C++ native · libn4m
pls4all.cpp.blas✓ ref0.10 ms
pls4all.cpp.blas+omp✓ ref0.11 ms0.11 ms
pls4all.cpp.omp✓ ref0.11 ms
pls4all.cpp.ref✓ ref0.10 ms0.15 ms
Python · pls4all
pls4all.python✓ bind0.08 ms🏆0.08 ms🏆
pls4all.sklearn✓ bind0.08 ms0.08 ms

See also: methods index · interactive dashboard